UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Operando Characterization of Intermediates Produced in a Lithium-Sulfur Battery

Gorlin, Y.; Siebel, A.; Piana, M.; Huthwelker, T.; Jha, H.; Monsch, G.; Kraus, F.; Gasteiger,H.A.; Tromp, M.Published in:Journal of the Electrochemical Society

DOI:10.1149/2.0081507jes

Link to publication

Citation for published version (APA):Gorlin, Y., Siebel, A., Piana, M., Huthwelker, T., Jha, H., Monsch, G., ... Tromp, M. (2015). OperandoCharacterization of Intermediates Produced in a Lithium-Sulfur Battery. Journal of the Electrochemical Society,162(7), A1146-A1155. https://doi.org/10.1149/2.0081507jes

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: https://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 05 Apr 2020

A1146 Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015)

Operando Characterization of Intermediates Producedin a Lithium-Sulfur BatteryYelena Gorlin,a,b,∗,z Armin Siebel,a,∗ Michele Piana,a,∗∗ Thomas Huthwelker,c

Himendra Jha,a,∗∗,f Georg Monsch,d Florian Kraus,d Hubert A. Gasteiger,a,∗∗∗and Moniek Trompb,e

aChair of Technical Electrochemistry, Department of Chemistry and Catalysis Research Center, Technische UniversitatMunchen, D-85747 Garching, GermanybCatalyst Characterization, Department of Chemistry and Catalysis Research Center, Technische UniversitatMunchen, D-85747 Garching, GermanycPaul Scherrer Institute, Swiss Light Source, Laboratory for Catalysis and Sustainable Chemistry, CH-5232 VilligenPSI, SwitzerlanddPhilipps-Universitat Marburg, Fachbereich Chemie, 35032 Marburg, GermanyeUniversity of Amsterdam, Van’t Hoff Institute for Molecular Sciences, 1098 XH Amsterdam, The Netherlands

One of the technological barriers to electrification of transport is the insufficient storage capacity of the Li-ion batteries on whichthe current electric cars are based. The lithium-sulfur (Li-S) battery is an advanced technology whose successful commercializationcan lead to significant gains in the storage capacity of batteries and promote wide-spread adoption of electric vehicles. Recently,important Li-S intermediates, including polysulfides, S3

•-, and Li2S, have been shown to present unique X-ray absorption nearedge structure (XANES) features at the sulfur K-edge. As a result, a combination of XANES characterization with electrochemistryhas the potential to contribute to the understanding of Li-S chemistry. In this study, we present an operando XANES cell design,benchmark its electrochemical and spectroscopic performance, and use it to track reaction intermediates during the discharge of thebattery. In particular, by employing electrolyte solvents with either a high or a low dielectric constant, we investigate the influence ofthe solvent on the conversion of polysulfide species to Li2S. Our results reveal that Li2S is already formed after ∼25–30% dischargein both types of electrolyte solvents, but that further conversion of polysulfides to Li2S proceeds more rapidly in a solvent with a lowdielectric constant.© The Author(s) 2015. Published by ECS. This is an open access article distributed under the terms of the Creative CommonsAttribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in anymedium, provided the original work is properly cited. [DOI: 10.1149/2.0081507jes] All rights reserved.

Manuscript submitted November 24, 2014; revised manuscript received March 2, 2015. Published March 27, 2015.

Electric vehicles represent a promising alternative to conventionaltransport based on an internal combustion engine and, when combinedwith renewable energy sources, have the potential to decrease world-wide CO2 emissions by 20–30%.1,2 In recent years, they have beengaining popularity in the form of passenger cars, with the number ofsold units surpassing the 100,000 mark in 2012.2 One of the maintechnological barriers to an even wider adoption of electric cars anda complete displacement of the CO2-emitting vehicles is the insuffi-cient storage capacity of the current battery technology, which leadsto a limited vehicle driving range.2,3 The useable energy density ofstate-of-the-art lithium-ion batteries employed in recent electric ve-hicles amounts to only ≈120 Wh/kgsystem,4 which corresponds to apractical driving range of 150–200 km. If one succeeds in the devel-opment of sufficiently durable high-voltage or high-energy cathodematerials and combines them with silicon anodes, an energy densityof ≈250 Wh/kgsystem would be in reach.4

An alternative promising battery concept is the lithium-sulfur (Li-S) battery.3,5–7 Current development in Li-S battery research includessuccessful commercialization of Li-S cells with 260 Wh/kgsystem en-ergy density by Sion Power corporation.8 This value already exceedsthe DOE target for 2022 (250 Wh/kgsystem)9 and has the potential toalso comply with the system cost target of 125 USD/kWh9 becauseof the abundance and low cost of sulfur. To achieve even greater gainsin the performance and, most importantly, to improve the cycle lifeof Li-S batteries, it will be necessary to gain a deeper mechanisticunderstanding of the 16 e–/S8 interconversion between S8 and Li2Sshown in reaction 1.10

S8 + 16 e− + 16 Li+ ⇀↽ 8 Li2 S [1]

Although the detailed intermediate (electro)chemical steps in-volved in reaction 1 have not yet been fully established, previous

∗Electrochemical Society Student Member.∗∗Electrochemical Society Active Member.

∗∗∗Electrochemical Society Fellow.fPresent address: Atotech Deutschland GmbH, 10553 Berlin, Germany.zE-mail: yelena.gorlin@tum.de

fundamental studies have identified that the initial discharge prod-ucts depend on the electrolyte solvent properties.11–14 In particu-lar, an additional intermediate characterized as a S3

•− (a radicalanion)15,16 has been shown to form in strongly ion-solvating envi-ronments (i.e., high dielectric solvents such as dimethyl sulfoxide,dimethylacetamide, and N,N-dimethylformamide), but not in poorlyion-solvating environments (i.e., low dielectric solvents such as 1,3-dioxolane, 1,2-dimethoxyethane, or their 1:1 mixture by volume).11,14

The formation of the radical likely proceeds according to the followingreactions:13–15,17

S8 + 2 e− → S2−8 [2]

S2−8 → S2−

6 + 1

4S8 [3]

S2−6

⇀↽ 2 S·−3 [4]

resulting in more than 2 e–/S8 consumed per S3•− generated (viz.,

2.67 e–/S8 based on reactions 2–4). The earliest support for thismechanism came from the study by Bonnaterre and Cauquis, whohad shown that the generated amount of S6

2−/S3•– peaks at around

2.7 e–/S8, before being subsequently reduced to S42-.15 In contrast

to the above mechanism, the initial reduction of S8 to S42– in low

dielectric solvents has been recently shown to proceed through a sin-gle reduction step in a rotating ring-disk electrode study.14 Similarmechanistic understanding of the subsequent conversion of S4

2− toLi2S through a variety of possible reduction, disproportionation, andprecipitation steps in different types of solvents has not yet been es-tablished and several competing mechanisms of Li2S formation havebeen proposed.

It was originally believed that S42- was first reduced to S2

2−/Li2S2

and the S22−/Li2S2 species were then subsequently reduced to

Li2S.6,18,19 Based on this sequential electro-reduction mechanism, itwas hypothesized that Li2S formation may be correlated with the finalchange in the slope of the galvanostatic discharge curve and the re-sulting incomplete utilization of sulfur.18 Several recent experimentaland theoretical studies, however, have demonstrated the possible for-

Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015) A1147

mation of Li2S early into the discharge, after only 4 e−/S8 consumed,instead of at the very end of the discharge.20–22 Such early detectionof Li2S suggests that the formation of lower order polysulfides andsolid products is simultaneous and not sequential.

The observed formation of Li2S after a consumption of only 4 e−/S8

could occur either through a series of concurrent electro-reductionsteps as proposed by Kumaresan and co-workers22 or through variousdisproportionation steps, as proposed by Lu and co-workers.14 Twoexamples of possible disproportionation steps, leading to formation ofLi2S and either a higher order polysulfide or S8 are shown in reactions5 and 6.

Li2 S4 ⇀↽4

7Li2 S + 3

7Li2 S8 [5]

Li2 S4 ⇀↽ Li2 S + 3

8S8 [6]

The main difficulty in the exact classification of the mechanismleading to the conversion of S4

2− to Li2S stems from the intractabil-ity of lower order polysulfides in cyclic voltammetry and potential-controlled half-cell studies, often referred to as short-term electrolysistests in literature.14,15,17 The reactions probed using these methods oc-cur on the timescale of seconds to minutes, while the conversion ofS4

2− to Li2S likely occurs on the timescale of minutes to hours.14 Fur-thermore, standard non-electrochemical characterization techniquesemployed to differentiate between various reduction products dur-ing long-term half-cell tests usually can detect either solid and crys-talline Li2S or liquid polysulfides, but not both solid and liquid prod-ucts at the same time. For example, X-ray diffraction (XRD)19,20 andsolid-state magic angle spinning nuclear magnetic resonance (NMR)spectroscopy23 have been previously used to successfully track Li2Sformation, while ultraviolet-visible (UV-Vis) spectroscopy,15,17,24,25

high performance liquid chromatography (HPLC),25 and solutionNMR23,26 have been used to identify polysulfide intermediates inthe electrolyte. Although some attempts have been made to extendUV-Vis spectroscopy and solution NMR techniques toward quantifi-cation of Li2S, these experiments saw only limited success, becauseLi2S lacked unique spectral features and had poor solubility in theelectrolyte.23,24,26 Consequently, new characterization techniques arenecessary to elucidate the process of Li2S formation during the dis-charge of Li-S batteries.

Recently, several Li-S reaction intermediates, includingpolysulfides,27–29 S3

•−,30,31 and Li2S solid product,27,32 have beenshown to present unique features in the sulfur K-edge X-ray ab-sorption near edge structure (XANES) spectrum. As a result, sulfurK-edge XANES has the potential to contribute to the understand-ing of the discharge mechanism in the 4–16 e–/S8 discharge region,which is known as the low voltage plateau, and facilitate the rationaldesign of improved Li-S batteries. Previous operando XANES stud-ies have employed a pouch or a coin cell, which had been modifiedby an X-ray transparent window.27,28,32–35 Such a cell design resultedin XANES spectra, which were completely governed by the sulfurelectrode and the electrolyte present in its pores, because of the lowpenetration depth of the X-rays at the sulfur K-edge energy. The restof the species, which were located in the electrolyte between the twoelectrodes, were almost invisible to spectroscopic characterization.28

In our work, we introduce an alternative cell design, which elimi-nates the back side of the sulfur electrode from the path of the X-raysand enables spectroscopic characterization of all three components ofthe battery: the cathode, the anode, and the separator. In the exper-iments, we first benchmark the electrochemical performance of ourcell against a standard Swagelok T-cell and then validate the abilityof our cell to measure intermediates by tracking the formation of S3

•−

in a cell containing an electrolyte solvent with strong ion-solvatingproperties. Having established the spectro-electrochemical capabili-ties of our operando cell, we use it to characterize the low-voltageplateau intermediates in two cells containing electrolyte solvents withsignificantly different dielectric constants. We successfully detect theinitial Li2S formation after a discharge of ∼4–5 e–/S8 in both types

Figure 1. a) Exploded view of the spectro-electrochemical operando cell, de-picting its individual components: 1) cell base; 2) X-ray window, aluminizedKapton; 3 and 4) aluminum front-plate and back-plate; 5) stainless-steel currentcollectors; 6) electrode-separator assembly; 7) stainless-steel electrical contactpins. b) Schematic of XANES setup illustrating monochromatic X-rays pen-etrating the cell through an aluminized Kapton window. XANES spectra aremeasured in fluorescence mode at 90◦ with respect to the incoming beam.Inset graphic shows a close-up of the electrode arrangement and measurementgeometry inside the electrochemical cell (electrical contact not shown); theX-ray beam can, in principle, be micro-focused on the cathode electrode, theseparator, and the anode electrode.

of electrolyte solvents and identify the influence of the solvent on thesubsequent transformation of polysulfides to Li2S.

Experimental Section

Cell design.— The operando cell design is presented in Figure 1a.The cell consists of a cell base (Fig. 1a, (1)) made from polyether etherketone (PEEK), an electrical insulator, which offers good mechanicalproperties and is chemically inert. An 8 micron thick Kapton foil met-allized with a 100 nm thick layer of aluminum (Multek, USA) is usedas an X-ray window (Fig. 1a, (2)) during the XANES experiments.It proves to be an excellent air and water barrier on the time scale ofhours, while offering about 50% transmission for X-rays at the sulfurK-edge at 2500 eV, when the cell is mounted at 45◦ relative to thebeam. The foil is positioned in between the cell base and an aluminumfront-plate (Fig. 1a, (3)). The front-plate has a 2 mm diameter hole,which is tapered by 50◦ to allow the X-rays to enter and leave the cell.Sealing of the electrolyte compartment is realized using two chem-ically inert Viton O-rings between the front-plate and an aluminumback-plate (Fig. 1a, (4)).

The cell base features a rectangular cut-out, open on both sides,in which the stainless-steel (DIN 1.4751) current collectors (Fig. 1a,(5)) and the electrode-separator assembly (Fig. 1a, (6)) reside. Theelectrical contact with the potentiostat is made through contact pins(Fig. 1a, (7)) on either side of the cell. The contact pins are compress-

A1148 Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015)

ing the current collectors with a constant force of ca. 20 N via twostainless-steel springs.

A schematic of the operando XANES setup demonstrating themeasurement geometry and cell components is shown in Figure 1b.Monochromatic, focused X-rays are entering the cell through the holein the front window and are passing through the aluminized Kaptonfoil. Inside the cell, they interact with sulfur species, resulting in theemission of fluorescent X-rays. After the fluorescent X-rays leave thecell, they are registered by an energy dispersive fluorescence detectorpositioned at an angle of 90◦ with respect to the incident beam. Wecalculate that we can measure the X-rays that pass through approx-imately 10–20 μm of the electrolyte. Because the cathode and theseparator are highly porous materials, they have a negligible contribu-tion toward attenuation of the X-rays. Although in this study we havemeasured with a defocused beam across the sulfur electrode and theseparator to minimize beam damage, the described cell design would,in principle, be able to conduct spatially-resolved XANES measure-ments of the cathode electrode, the separator, and the anode electrode.In these measurements the ratio of electrolyte to sulfur loading willhave to be carefully considered, as it will have an influence on theequilibrium of the species that form inside the separator.

Electrochemical measurements.— Electrochemical measurementswere performed in a standard stainless-steel T-cell made fromSwagelok components and in the operando cell. Electrolytes wereprepared from 1 M lithium perchlorate (LiClO4, battery grade,99.99% trace metal basis, Sigma-Aldrich) and 0.5 M lithium ni-trate (LiNO3, 99.99% trace metal basis, Sigma-Aldrich), an addi-tive to enhance surface passivation of the Li electrode, dissolved ineither a mixture of 1,3-dioxolane (DOL, anhydrous, 99.8%, Sigma-Aldrich) and 1,2-dimethoxyethane (DME, anhydrous, 99.8%, Sigma-Aldrich) (1:1 vol:vol) or N,N-dimethylacetamide (DMAC, anhydrous,99.8%, Sigma-Aldrich). These solvents will be subsequently referredto as DOL-DME and DMAC, respectively. We chose LiClO4 as anelectrolyte salt rather than the commonly used bis(trifluoromethane-sulfonyl)imide lithium (LiTFSI) in order to eliminate the additionalcontribution of sulfonyl group to the XANES spectrum. Preliminaryexperiments were also performed using 1 M LiTFSI (99.95 % tracemetal basis, Aldrich) and 0.5 M LiNO3 dissolved in DOL-DME inboth the Swagelok T-cell and the operando cell to benchmark the per-formance of the operando cell and to compare the discharge behaviorof the battery in the electrolytes based on LiClO4 or LiTFSI salts.

The sulfur/carbon (S8/C) composite was prepared by wet impreg-nation of 1.5 g Vulcan carbon (XC-72, Tanaka Kikinzoku Kogyo) with120 ml of a 1 M solution of sodium thiosulfate (Na2S2O3 · 5 H2O,99.5%, Sigma-Aldrich) for 10 min.36 The wet impregnation step wasfollowed by 20 min ultrasonic treatment after which a well-dispersed,homogeneous suspension was obtained. To this suspension an excessof 250 ml of 1 M HNO3 (ACS reagent, Sigma-Aldrich) was added,and the mixture was sonicated for another 20 min. Afterwards, thesupernatant was decanted, washed three times with double-distilledwater, and filtered to obtain the composite. The obtained compositewas dried over silica gel in a sealed glass oven (Buchi, Switzerland) at80◦C for 72 h. Finally, the composition of the material was determinedto be 67 wt% sulfur by thermogravimetric analysis.

To prepare an electrode, the S8/C composite was added to a sus-pension of polyvinylidene difluoride (PVDF HSV900, Kynar) in N-Methyl-pyrrolidinone (NMP, 99.5%, anhydrous, Sigma-Aldrich). Theweight of the composite and PVDF were adjusted to give an ink com-position of 85 wt% S8/C composite and 15 wt% PVDF. The resultingink was vigorously mixed for 2 h, spread onto a current collector(17 μm thick aluminum foil) using a Mayer rod with a wet-filmthickness of 250 μm, and dried for 72 h. These three steps wereall performed at room temperature. From the resulting coating, thecathode electrodes were punched out and determined to have a sulfurloading of 1.8 ± 0.1 mg/cm2

electrode.The electrochemical cells were assembled using these electrodes

at the cathode and a piece of lithium metal (99.9% purity, Rock-wood Lithium, USA) at the anode. The area of the electrodes was

either 0.80 cm2 or 0.90 cm2, depending on whether they were usedwith the Swagelok T-cell or the operando cell, respectively. For theexperiments in DMAC electrolyte, lithium was pre-conditioned for24 h in propylene carbonate (PC, 99.7% purity, Aldrich) contain-ing 1 M LiClO4. Sulfur and lithium electrodes were separated bya 200 μm glass-fiber separator (Glass microfiber filter 691, VWR,Germany), which was soaked with 100 μl of the electrolyte. Dueto the high sensitivity of the cell chemistry and, particularly, of thelithium metal, to air and water, all components, with the exception ofS8/C electrodes, were thoroughly dried at 70◦C under dynamic vac-uum for a minimum of 1 h before being transferred into a glove boxcontaining less than 1 ppm of oxygen and water. S8/C electrodes weredried over silica gel at 80◦C to avoid sublimation of sulfur.

The assembled electrochemical cells were connected to a SP-200single-channel potentiostat (Bio-Logic SAS, France), which was firstused to record a potentio-electrochemical impedance spectrum andthen to galvanostatically discharge the cell at a C-rate of 0.1 h–1 (basedon a theoretical capacity of 1672 mAh/gs) to a cut-off voltage of 1.5 Vvs. Li/Li+. The open circuit voltage (OCV) period between the cellassembly and the beginning of the discharge was always 1–2 h.

Sulfur K-edge XANES measurements.— Sulfur K-edge XANESmeasurements were performed at the PHOENIX beamline of theSwiss Light Source (SLS, Paul Scherrer Institut, Villigen, Switzer-land) and at I18 beamline of the Diamond Light Source (DLS, Didcot,UK). The initial cell was commissioned and tested at I18 beamlineof the DLS, while the actual operando data, which is presented inthis paper, were obtained at the PHOENIX beamline of the SLS. ThePHOENIX beamline is a tender X-ray beamline, covering the energyrange from 0.8 to 8 keV. Its optical concept closely follows the LUCIAbeamline at Soleil synchrotron facility.37 In its setup, an elliptical un-dulator serves as source for the X-rays. Monochromatic X-rays withan energy resolution of about 0.3 eV are generated using a Si(111)double-crystal monochromator (DCM). The beam is subsequently fo-cused using a set of mirrors in Kirkpatrick-Baez configuration, withthe maximum possible focus corresponding to a spot size of 2.5 μm× 2.5 μm. In the experiments, the operando cell was mounted in alarge sample chamber filled with helium gas at a pressure of 750 mbar.XANES spectra were recorded at the sulfur K-edge in fluorescencemode using a 4 element Vortex detector. Data acquisition was runcontinuously yielding one spectrum every 8 minutes for the DOL-DME cell and one spectrum every 9 minutes for the DMAC cell.These acquisition times corresponded to an electrochemical chargeincrement equivalent to 0.22 and 0.24 e–/S8 (23 and 25 mAh/gs), re-spectively. Fluorescence XANES spectra were normalized using thefluorescence signal of the aluminum window (Al Kα Fluorescence)as I0. The acquired XANES spectra were then processed using theAthena software package.38

Before performing the operando experiments, we collectedXANES spectra of an S8/C composite electrode with 100 μl of theDOL-DME electrolyte containing 1 M LiClO4 and 0.5 M LiNO3 in awell-sealed container. As expected, in the first spectrum, only the sig-nature of elemental sulfur was observed. Subsequent spectra recordedin the same spot, however, demonstrated an emergence of two newfeatures at incident energies of 2482 eV and 2498 eV. The fact thatthese features were not present in the initial spectrum and that noelectro-chemical reaction could occur inside the sealed cell, suggeststhat their appearance is linked to beam-induced oxidation of sulfur inthe presence of the electrolyte. Comparison of the spectra with thoseof Li2SO4 powder identifies the produced species as sulfate (Supple-mental material, Fig. S1a). The presence of beam-induced oxidationis further substantiated by the fact that the concentration of sulfate(expressed as the magnitude of the peak) is related to irradiation timeand that a clean spectrum of elemental sulfur can be obtained aftermoving the beam to a new spot on the sample. To reduce the extentof radiation damage we defocused the beam from its initial spot sizeof 10 μm x 10 μm to a spot size of 1800 μm x 50 μm and re-duced the flux of the incoming photons by one order of magnitude bychanging to a higher harmonic of the undulator. The combination of

Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015) A1149

these measures, which resulted in a ∼104-fold lower radiation flux,successfully avoided radiation-induced reactions (Supplemental ma-terial, Fig. S1b) and, ultimately, allowed continuous acquisition ofXANES spectra over the course of ∼7 hours. After the dischargehad lasted longer than 7 hours in the DOL-DME cell, a small beam-induced feature was observed (Fig. 6), although no correspondingbeam-damage was observed in the DMAC cell. One disadvantage ofusing a defocused beam was that it made it difficult to selectivelymeasure the signal from either the sulfur electrode or the separator,and the XANES data presented in this study corresponds to a signalaverage over these two components of the battery. In our future work,we will optimize the thicknesses of the electrodes and the separator, aswell as beamline configuration, to allow spatially-resolved XANESmeasurements in the absence of beam-induced oxidation of sulfurintermediates.

Data analysis.— Qualitative analysis of XANES spectra was per-formed using 3 powder standards: S8, Na2S4, and Li2S. Commerciallyavailable powders were used for S8 (99.998% trace metal basis, Sigma-Aldrich) and Li2S (99.98% trace metal basis, Sigma-Aldrich). Priorto XANES characterization the powders were manually ground anddiluted with cellulose to a final concentration of 10 wt%. Na2S4 poly-sulfide was prepared using a known literature procedure.39 Briefly, astoichiometric mixture of Na metal (1 mmol) and sulfur was added toapproximately 0.5 ml of anhydrous ammonia in a flame-sealed bombtube at room temperature. After a reaction time of a couple of days,the bomb tubes were cooled to −78◦C and opened, the ammonia waspumped off, and Na2S4 powder was obtained. The phase purity of thepowder was checked using powder-XRD and infrared spectroscopy.Na2S4 was subsequently diluted following the same procedure as S8

and Li2S powder standards. Although we also attempted to synthesizeLi2S4 powder, it could not be isolated, because it is thermodynamicallyunstable.27,40 Since the XANES features of Na2S4(27, 28) are simi-lar to the predicted features of lithium polysulfides,29 it is a suitablereference for a qualitative analysis.

Quantitative analysis of XANES data was attempted using a linearcombination analysis procedure, but it was not possible to obtaina good fit for all the spectra due to distorted XANES features inmany of our operando and reference spectra. These distortions can beexplained by self-absorption of the X-rays, a common problem seenwith sulfur samples.32,41–43 It is not possible to correct the distortions inoperando spectra, in which neither the species nor their concentrationsare known,28,43 and therefore, no self-absorption corrections wereapplied. In principle, self-absorption effects can be eliminated bydiluting the amount of sulfur in the sample and reducing the size ofthe sulfur particles below 1 μm.43 We illustrate this possibility bycharacterizing a dilution series of sub-micrometer sized sulfur in ourS8/C composite and a dilute liquid sample of sulfur in the DOL-DMEsolvent (4 mM S8 or 32 mM sulfur). The resulting spectra of the initialcomposite and the composite samples diluted with Vulcan carbon to10, 5, 1, 0.5, and 0.1 wt% of sulfur are presented in Figure 2a, whilethe dilute liquid sample of sulfur is compared to 0.1 wt% S8/C samplein Figure 2b. The figures demonstrate that the sulfur XANES featureremains distorted when the concentration of sulfur is at 1 wt% orhigher. Since the concentration of sulfur in carbon-based compositesor electrodes is always much higher than 1 wt% in Li-S batteries,and since the polysulfide equilibria are known to be concentrationdependent,11 it was not practical to aim to completely avoid self-absorption effects in the operando experiments. Such efforts wouldlead to a system, which would not be representative of a real batterycell. Consequently, we used a typical sulfur composite with 67 wt%of sulfur and focused on the qualitative interpretation of the operandoXANES measurements.

Results and Discussion

Benchmarking of electrochemical performance.— The electro-chemical performance of the operando cell was benchmarked againsta Swagelok T-cell in a standard electrolyte, consisting of 1 M LiTFSI

Figure 2. a) Normalized sulfur K-edge XANES spectra of the as-preparedS8/C composite (67 wt% sulfur) and S8/C composite diluted with Vulcan car-bon to 10, 5, 1, 0.5, and 0.1 wt% of sulfur. Distortions in the spectrum arepresent at 1 wt% sulfur and higher concentrations; b) XANES spectrum of4 mM S8 dissolved in DOL-DME compared to the spectrum of S8/C com-posite diluted to 0.1 wt% of sulfur. Neither sample contains self-absorptiondistortions.

and 0.5 M LiNO3 dissolved in DOL-DME (1:1 vol:vol), using dis-charge and charge current of 167 mA/gs (i.e., a C-rate of 0.1 h-1).Owing to the identical electrode configuration in the Swagelok cell(red) and in our operando cell (blue), the discharge and chargecurves are similar in both cells (Figure 3a). Although only one rep-resentative curve is shown, the operando cell could reproduciblyachieve a first discharge capacity of ca. 1100–1300 mAh/gS andsubsequently charge, thus matching the performance of a stan-dard Li-S cell and approaching the theoretical capacity of Li-Schemistry.

Additional benchmarking experiments were also performed in twoother electrolytes, which were subsequently used in the operandoXANES experiments: 1 M LiClO4 and 0.5 M LiNO3 dissolved inDOL-DME and 1 M LiClO4 and 0.5 M LiNO3 dissolved in DMAC.Because operando experiments focused on the discharge mechanism,only the discharge curves are shown in Figures 3b and 3c. The twofigures demonstrate that during XANES characterization the replace-ment of LiTFSI with LiClO4 does not have a significant effect on thereduction potentials of the high and low voltage plateaus, while thereplacement of the low dielectric DOL-DME solvent with the high di-electric DMAC solvent results in the predicted shifts in the OCV andthe reduction potentials due to substantially different Li+ solvationenergies in the two types of electrolyte solvents.14,44

The successful electrochemical benchmarking of our operandocell indicates that it has unique capabilities. Other published or com-mercially available spectro-electrochemical cell designs are unableto achieve both standard electrochemical performance and make allthree components of the cell available to spectroscopic characteri-

A1150 Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015)

Figure 3. a) Comparison of discharge and charge curves at a C-rate of0.1 h–1 obtained in the T-cell and in the operando cell in a standard elec-trolyte, consisting of 1 M LiTFSI and 0.5 M LiNO3 dissolved in DOL-DMEmixture. b), c) Comparison of the discharge curves in three different elec-trolytes (1 M LiTFSI and 0.5 M LiNO3 dissolved in DOL-DME, 1 M LiClO4and 0.5 M LiNO3 dissolved in DOL-DME, and 1 M LiClO4 and 0.5 M LiNO3dissolved in DMAC) in the operando cell b) and the T-cell c).

zation. The most common cell design employed with Raman, X-rayphoto-electron and XANES spectroscopies relies on a modification toa current collector of a coin, pouch, or Swagelok cell through an intro-duction of a window compatible with the desired technique.27,28,45,46

In this setup, the excellent electrochemical performance of the cell ismaintained, but it is only possible to measure a sum signal of the elec-trode and the electrolyte found in the pores of the electrode. Anothercell design has followed a different strategy by pressing the entiresurface area of both electrodes, in a side-by-side arrangement, againstthe window material.47,48 Such design enables spectroscopic access toeach of the two electrodes and the separator, but only at the expenseof poor electrochemical performance.47 Hence, the capability of thepresented operando cell design to maintain a favorable electrode con-figuration and therefore, standard electrochemical performance, whilesimultaneously exposing the two electrodes and the separator to theX-rays, is a significant improvement over the commercially availableand previously published spectro-electrochemical cell designs.

Validation of operando characterization of intermediates.— Theoperando discharge curves for the DMAC and DOL-DME cells arepresented in Figures 4a and 4c, both as a function of cell capacity andof the number of consumed e−/S8. They demonstrate the expectedhigh and low voltage plateaus and typical discharge capacities of 12–13 e–/S8 (1250–1350 mAh/gs). The operando XANES spectra werecollected continuously, with each spectrum lasting either 9 (DMAC)or 8 (DOL-DME) minutes, which corresponded to a discharge intervalof 0.24 e–/S8 (25 mAh/gs) in DMAC and 0.22 e–/S8 (23 mAh/gs) inDOL-DME. To verify that our cell design could provide accurateinformation on the composition of intermediates, we first focused onthe XANES spectra obtained in the well characterized 0–4 e–/S8 initialdischarge region.

To interpret the XANES spectra we relied on three standards: S8,S4

2− polysulfide, and Li2S. The XANES spectra of S8 and Li2S weremeasured using commercially available powders, while the spectrumof S4

2− polysulfide was approximated by the spectrum of a synthe-sized Na2S4 powder. Although the spectra of other polysulfides such asS3

2−, S52−, S6

2−, S82− are not presented in this study, the positions of

their XANES features are expected to be well represented by the spec-trum of S4

2− polysulfide.27–29 This is rationalized by a computationalstudy, which has demonstrated that the charge on a polysulfide chainmigrates toward the two terminal sulfur atoms.29 This phenomenonleads to a sulfur K-edge XANES signal consisting of two features,with the first feature at ∼2470–1 eV corresponding to the negativelycharged terminal sulfur atoms and the second feature at ∼2472–3eV corresponding to the uncharged (‘neutral’) internal sulfur atoms.As the length of the polysulfide chain decreases, the absolute energyposition of each feature experiences little change, while the intensityratio between the second and the first peak decreases. The spectra ofvarious polysulfides can thus be differentiated by the relative inten-sities or areas of the two peaks.29 During operando characterization,however, S8 and many different polysulfides can be present simultane-ously, which makes quantitative analysis of the relative concentrationsof the polysulfides difficult, if not impossible.

The spectra of the references are shown in Figure 5 and demon-strate the expected characteristics of S8,46 polysulfides,29 and Li2S.27,29

Specifically, S8 has a single major peak at 2472 eV, S42− has two ma-

jor peaks at 2470 eV and 2472 eV, and Li2S has two major peaks at2473 eV and 2475.5 eV and an additional feature at 2483 eV. AlthoughLi2S is in a more reduced state than S8 and polysulfides, the presenceof its main features in the 2473–2475.5 eV region can be explainedby the ionic character of the Li2S bond.49 In our qualitative analysiswe will identify formation of polysulfides by the appearance of the2470 eV feature and the formation of Li2S by the appearance of aconvex feature at 2475.5 eV (Figure 5). The slightly dampened shapeof all the peaks indicates the presence of self-absorption, which can beexplained by the use of manual grinding procedure and the resultingmicrometer-sized and not the needed sub-micrometer-sized particlesand/or slightly too high amount of the sulfur compound used.41 In theinset of the figure, we also include the spectrum of 4 mM S8 in DOL-DME to compare a sample without any self-absorption distortionsto our three standards. The observed presence of the self-absorptionartifact does not restrict qualitative interpretation of the operandoXANES spectra, which, as has been previously explained, also expe-riences self-absorption problems.

Figures 4b and 4d show the operando XANES spectra collectedimmediately prior to the beginning of the discharge (0 e−) and af-ter a consumption of 1, 2, 3, and 4 e–/S8 in reaction 1. From theXANES spectra corresponding to 0 e–/S8, which are characterized bya major peak at an incident X-ray energy of 2472 eV and match theself-absorbed spectrum of the S8 reference in Figure 5, it can be con-cluded that only S8 is present in the DMAC and DOL-DME operandocells after the initial OCV period of 1.5–2 hours. As the dischargeproceeds, a new feature appears at an incident energy that is approxi-mately 2 eV lower than the main feature of S8, indicating reduction ofS8 to polysulfides. Subsequently, the absorption intensity of the newlyacquired feature grows as the number of the transferred electrons in-creases from 1 e−/S8 to 4 e−/S8, while, at the same time, the absorption

Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015) A1151

Figure 4. Operando discharge curve in a) DMAC and c) DOL-DME cells plotted both as a function of the number of electrons consumed in S8 + 16 Li +16 e− → 8 Li2S reaction and as a function of specific capacity (at a C-rate of 0.1 h–1). Normalized operando sulfur K-edge XANES spectra obtained in thewell-characterized 0–4 e–/S8

− discharge region of the b) DMAC and d) DOL-DME cells, with the inset graphics visualizing the ratio between the two polysulfidespeaks at 2472 eV and 2470 eV. To estimate the peak intensity of the feature at 2470 eV, a background subtraction was performed using the normalized absorptionvalues of the 0 e−/S8 spectrum in the 2469–2471 eV region and the resulting value at 2469.9 eV was used as the peak intensity. The intensity of the peak at 2472 eVcorresponds to the normalized absorption at 2472 eV.

intensity of the main feature at 2472 eV either decreases or remainsconstant. These changes are visualized in the insets of Figures 4b and4d by plotting the ratio of the peak intensities located at 2472 eV and2470 eV as a function of transferred electrons (1–4 e–/S8).

As discussed in the introduction, the discharge mechanism inthe 0–4 e–/S8 region in the high dielectric DMAC electrolyte is ex-pected to be different from the discharge mechanism in the low di-electric DOL-DME electrolyte. In particular, previous UV-Vis spec-troscopy studies have shown that the S3

•− appears in high but not lowdielectric solvents,11 and that its concentration reaches a maximumafter a charge transfer of 2.7 e−/S8.15 Recently, a unique XANES fea-ture of the S3

•− has been identified by Fleet and Liu30 and confirmedby Cuisinier et al.31 This feature corresponds to 1s → 3p(π∗) tran-sition of the unpaired electron and manifests as an additional signalin the 2468–2469 eV region of a typical polysulfide spectrum.30 Aclose inspection of the XANES spectra in Figure 4 reveals that anadditional shoulder is present in the spectra of the DMAC operandocell at a discharge of 3 e–/S8. Consequently, we used the observedfeature of the S3

•− to assess the ability of our operando cell to trackpolysulfide intermediates.

To compare our operando results to the previous UV-Vis spec-troscopy studies, we examined the signal at 2468–2469 eV in bothDMAC and DOL-DME in the 2.0–4.2 e–/S8 region of the dischargecurves (Fig. 6). In the DMAC cell (Figs. 6a and S2) a new XANES fea-ture, which manifests as a shoulder at ∼2468.4 eV, becomes prominentafter a charge transfer of 2.3 e–/S8. This feature subsequently reaches

a maximum after a charge transfer of 2.5–2.8 e–/S8 and then decreasesgradually, until 4.2 e−/S8 are transferred. These results closely matchthe UV-Vis spectroscopy results of Bonnaterre and Cauquis15 and arein excellent agreement with the initial reduction mechanism of S8

summarized by reactions 2–4, but they disagree with the recent workof Cuisinier et al., which claims that the intensity of S3

•− signal peaksat 3.2 e−/S8 (340 mAh/gs) and persists until 8.1 e−/S8 (850 mAh/gs)in the same electrolyte.31 The differences in the development of S3

•−

intensities presented in this work and in the work of Cuisinier et al. canbe explained by an interruption to the discharge that occurred in thework of Cuisinier et al.31 The interruption occurred after a dischargeof ∼2.7 e−/S8 and lasted 2 hours. Although no additional dischargeoccurred in this period, the disproportionation reactions continued totake place, which must have led to significant changes in the rela-tive concentrations of the intermediates. These considerations explainwhy the intensities of the XANES features reported by Cuisinier et al.are not representative of the intermediates that form during a standarddischarge and cannot be reliably associated with a particular dischargecapacity. In the DOL-DME cell (Figs. 6b and S2), the spectra are fea-tureless in the same 2468–2469 eV region, indicating the lack of S3

•−

formation in the low dielectric solvent. This result is consistent withthe absence of features in the 2468–2469 eV region in all previousoperando XANES studies that used low dielectric solvents (e.g., inDOL-DME27,31,32,34 and TEGDME,35 even though the authors of thelatter study incorrectly interpreted the feature at 2470 eV to correspondto the S3

•−). The successful observation of S3•− intermediate in high

A1152 Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015)

Figure 5. Normalized sulfur K-edge spectra of three standards: S8,S4

2−polysulfide, and Li2S. The spectrum of S42− polysulfide was obtained us-

ing Na2S4 powder. To clearly display the differences in the features of the threespecies the spectra of S4

2− polysulfide and Li2S are offset by 0.7 and 1.4 units,respectively. Because the standards demonstrate some self-absorption distor-tions, the inset compares the undistorted spectrum of 4 mM S8 in DOL-DMEto the spectra of the three standards. In our analysis, formation of polysulfidesis identified by the appearance of the peak at ∼2470 eV, which corresponds tothe charged terminal sulfur atoms, while formation of Li2S product is identi-fied by the formation of a peak at ∼2475.5 eV. At 2475.5 eV the spectrum ofLi2S has a clear convex shape, the spectrum of the polysulfide does not have asignificant feature, and the spectrum of S8 has a concave shape.

but not low dielectric solvent at the predicted number of dischargedelectrons confirms the ability of our cell to reliably track polysul-fide intermediates and verifies the spectroscopic capabilities of ourcell.

Detection of Li2S formation in the low-voltage plateau.— Havingvalidated our experimental method in the well characterized 0–4 e–/S8

initial discharge region, we now extend it to the less characterized lowvoltage discharge plateau (4–12 e–/S8) to understand the dependenceof discharge products formation on the strength of the ion-solvationproperties of the environment. If the electro-reduction mechanismwere strictly sequential, after 8 e–/S8 are discharged in reaction 1,S8 would be entirely converted to Li2S2 intermediate (reaction 7).During the subsequent discharge steps, Li2S2 could be reduced toLi2S as shown in reaction 8.

S8 + 8 e− + 8 Li+ ⇀↽ 4 Li2 S2 [7]

4 Li2 S2 + 8 e− + 8 Li+ ⇀↽ 8 Li2 S [8]

Alternatively, during a simultaneous electro-reduction process,which allows reactions 7 and 8 to occur in parallel,22 or as a re-sult of a disproportionation mechanism proposed by Lu et al.,14 Li2Scould form earlier in the discharge, before 8 e–/S8 are consumed.To understand how the nature of the intermediates changes during thedischarge in the low voltage plateau, we examine the spectra that werecollected in DMAC and DOL-DME cells, as the extent of dischargeincreased from 4 e–/S8 to 12 e–/S8. We note that because the appliedpotential in the 4–12 e–/S8 discharge region is below 2.1 V vs. Li in theDOL-DME cell and 1.8 V in the DMAC cell, we do not consider S8

reduction as a possible contribution to the changes seen in the XANESfeature at 2472 eV. At these potentials S8 would either rapidly reduce,as was previously demonstrated using RRDE characterization,14 orremain present as electrically isolated species.

Figure 6. a) Normalized operando sulfur K-edge XANES spectra demon-strating the observation of S3

•− intermediate in the 2.0–3.9 e–/S8 region ofthe DMAC cell with an inset graphic visualizing the peak intensity of theS3

•− feature. The peak intensity was determined by subtracting the spectrumcollected after a discharge of 1.3 e–/S8, in which no feature was present in the2468–2469 eV region, and reporting the peak maximum of each spectrum. b)Normalized operando sulfur K-edge XANES spectra demonstrating the lack offormation of S3

•− intermediate in the 2.1–4.1 e–/S8 region of the DOL-DMEcell.

Figure 7a shows five spectra corresponding to discharge states af-ter 4, 6, 8, 10, and 12 e–/S8 are transferred in the DMAC cell. All fivespectra contain the polysulfide peak at 2470 eV. This polysulfide sig-nature feature and its accompanying main peak at 2472 eV decreasein intensity as the extent of discharge increases, which indicates thatthe overall proportion of polysulfides inside the battery is decreas-ing. The Li2S feature at 2475.5 eV is prominent after a discharge of6–12 e–/S8, but a small convex peak can already be seen after a dis-charge of 4 e–/S8. To clarify the onset of Li2S formation, we examinefive operando spectra collected in the DMAC electrolyte solvent inthe 2–6 e–/S8 discharge region in Figure 7b. After a discharge of 2–3e–/S8 the shape of the feature at 2475.5 eV is still concave, indicatingthat little to no Li2S is present in the cell. As the extent of dischargeincreases to 4–5 e–/S8, however, the feature at 2475.5 eV becomes con-vex. This observed change in the peak shape provides direct evidencethat Li2S is present in the cell after a discharge of 4–5 e–/S8.

To determine if there is a difference in the observed onset of Li2Sformation between DMAC and DOL-DME cells, Figure 7c shows fiveoperando spectra corresponding to the 4–12 e–/S8 discharge regionof the DOL-DME cell. As in the DMAC cell, all spectra contain thepolysulfide feature at 2470 eV and a growing signal at 2475.5 eV.Unlike the DMAC cell, the main polysulfide peak at 2472 eV is lessprominent in the DOL-DME cell and after an electron transfer ofonly 8 e–/S8 decreases to a normalized absorption intensity value thatis below the value of the Li2S feature at 2475.5 eV. Additionally,two extra features are seen in the DOL-DME spectra at 2478 eVand 2482 eV. These features correspond to beam-induced oxidation

Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015) A1153

Figure 7. Examination of low voltage plateau intermediates using five normalized operando sulfur K-edge XANES spectra in a) DMAC and c) DOL-DME cellscorresponding to 4, 6, 8, 10, and 12 e–/S8. Identification of approximate onset of Li2S formation using five normalized operando sulfur K-edge XANES spectrain b) DMAC and d) DOL-DME cells corresponding to 2, 3, 4, 5, and 6 e–/S8. The figure demonstrates that the formation of Li2S begins at 4–5 e–/S8 in both typesof cells.

of sulfur species (Figure S1) and are not related to the dischargemechanism. The signal at 2475.5 eV has a clear contribution from Li2Sspectrum after an electrochemical transfer of 6–12 e–/S8, indicatingthat Li2S appears near the beginning of the low voltage plateau. Toclarify the onset of Li2S formation, Figure 7d examines five operandospectra collected in the DOL-DME cell in the 2–6 e–/S8 dischargeregion. In similarity to the spectra obtained in the DMAC cell, DOL-DME spectra collected after an electrochemical transfer of 2–3 e–/S8

have a concave shape at 2475.5 eV and thus, contain little to no Li2S.As the extent of discharge increases to 4 e–/S8 the feature transitionsfrom a concave to a convex shape, while a further increase in thenumber of electrons transferred to 5–6 e–/S8 leads to a formation ofa clear convex peak. These observed transformations in the shapeof the XANES feature at 2475.5 eV indicate that the onset of Li2Sformation in the DOL-DME cell occurs at a similar point of thedischarge as in the DMAC cell (4–5 e–/S8). The identified formationof Li2S after a discharge of 4–5 e–/S8 (420–520 mAh/gs) in bothcells confirms that Li2S coexists with polysulfides at the beginning ofthe low-voltage plateau, as has been previously suggested by XRD20

and electrochemical impedance spectroscopy,21 and provides directevidence against the sequential reduction mechanism, in which Li2Sis expected to form only after all of S8 is reduced to Li2S2.

Interestingly, previous operando XANES studies have not demon-strated Li2S formation at such an early point of the discharge usingstandard electrolyte and electrode formulations. In one study, whichreported both ex situ and in situ sulfur K-edge XANES spectra in asulfolane solvent with 1 M LiTFSI salt, formation of Li2S was de-tected only at the very end of the discharge.28 In a different study,which used a DOL-DME solvent with 1 M LiClO4 and 0.5 M LiNO3,Li2S was found to appear in the middle of the low-voltage plateau ata discharge capacity of 700–800 mAh/gs (6.7–7.7 e–/S8).27 Finally, intwo very recent studies, which characterized the discharge behaviorof both a standard Li-S battery based on the DOL-DME electrolytesolvent and S8/C composite and a battery employing either a newelectrolyte32 or a new type of electrode,34 Li2S formation was demon-

strated to occur as early as 250–320 mAh/gs (2.3–3.0 e–/S8) in thenovel systems, but a lack of Li2S was reported at the beginning of thelow-voltage plateau in the standard cells.32,34 Specifically, the authorseither demonstrated an absence of Li2S contribution to sulfur K-edgeXANES at 450 mAh/gs

34 or detected an onset of Li2S formation at1000 mAh/gs.32 While the exact reason for the higher sensitivity ofour operando cell to the detection of Li2S is not known, it may berelated to its unique measurement geometry.

Having established that the onset of Li2S formation occurs at thebeginning of the low-voltage plateau in both cells, we examine indetail the observed differences in the spectra of the two cells after anelectrochemical transfer of 8 e–/S8. Figure 8 directly compares thespectra obtained in the DMAC and DOL-DME cells after a discharge

Figure 8. Comparison of the normalized operando sulfur K-edge XANESspectra in DMAC and DOL-DME cells collected after an electrochemicaltransfer of 8, 10, and 12 e–/S8. The spectra corresponding to a discharge of 10and 12 e–/S8 are offset by 1.0 and 2.0 units, respectively.

A1154 Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015)

Figure 9. a) Comparison of Li2S and 4 mM S8 reference spectra to the litera-ture results from Ref. 27 and the normalized operando sulfur K-edge XANESspectra collected at the very end of discharge in DMAC and DOL-DME cells.To facilitate comparison of the spectra, the incident X-ray energy of the spec-trum from ref. 27 is calibrated to the same energy scale by shifting it to a higherX-ray energy by 2.2 eV, the spectrum of 4 mM S8 is multiplied by 0.25, andthe spectra of 4 mM S8, DOL-DME, DMAC, and ref. 27 are offset by 0.4, 0.8,1.2, and 1.6 units, respectively. The extent of the electrochemical dischargecorresponds to 11.4 e–/S8 in ref. 27 and to 11.8 e–/S8 and 12.6 e–/S8 in ourDMAC and DOL-DME cells. b) Normalized sulfur K-edge XANES spectracorresponding to the difference between the last spectrum obtained in the twocells and the Li2S reference spectrum.

of 8, 10, and 12 e–/S8. It clearly demonstrates that at a discharge stateof 8 e–/S8 the Li2S feature at 2475.5 eV is the main feature in the DOL-DME cell, while the polysulfide features at 2472 eV and 2470 eV arethe main features in the DMAC cell. As the number of transferredelectrons increases to 10 and then 12 e–/S8, the normalized absorptionintensity in the 2470–2472 eV region continues to decrease, while theXANES signal from Li2S product becomes more prominently seenat 2473 eV, where a second Li2S peak is expected. In particular, inthe DMAC cell the polysulfide feature at 2472 eV decreases until itsintensity is about equal to the intensity of the peak at 2473 eV, while inthe DOL-DME cell the intensity of the polysulfide feature drops belowthe intensity of the peak at 2473 eV. These observed differences inthe DOL-DME and DMAC spectra in the 8–12 e–/S8 discharge regionindicate a smaller relative proportion of polysulfides in the DOL-DMEcell and suggest that the disproportionation of polysulfides in a poorlyion-solvating environment of DOL-DME solvent is more rapid thanin the strongly ion-solvating environment of DMAC solvent.

Previous operando XANES studies have demonstrated a surpris-ing presence of unreacted S8 at the end of the discharge of Li-Sbattery.27,28,33 To determine if our tested cells also have S8 at theend of discharge, which, for example, could have been electricallyisolated from the cathode and therefore, remained in an unreacted ox-idized form, Figure 9a compares the final spectra obtained in DMACand DOL-DME cells to the previous literature results of Cuisinier et

al.27 In disagreement with previous results, we do not see a significantfeature at 2472 eV in either cell, suggesting that when our operandocells reach the end of the discharge, little to no S8 is present. To betterunderstand the type of species that are present in the last spectra, wesubtract Li2S features from each spectrum and visualize the differencespectra in Figure 9b. Inspection of the subtracted XANES spectrumof the DMAC cell (blue spectrum in Fig. 9b) reveals the presenceof polysulfide species, while inspection of the subtracted XANESspectrum of the DOL-DME cell (red spectrum in Fig. 9b) reveals acomponent that does not correspond to a typical polysulfide spectrum,but rather has a single broad feature at 2470.5 eV. The presence of aspecies with a feature at 2470.5 eV has also been previously detectedin an ex situ XANES spectrum by Patel et al., who have postulatedthat it corresponds to a different form of Li2S.28 We hypothesize thatthe feature corresponded not to Li2S but to S2

2-/Li2S2, based on thefirst-principles calculations of Pascal et al.,29 who have shown thatS2

2-/Li2S2 is expected to have a lone peak at an energy that is approx-imately 0.5 eV more positive of the original polysulfide feature.

The observed differences in the final composition of species inDMAC and DOL-DME cells suggest that different mechanisms maycontribute to reaching the end of discharge in the two different ion-solvating environments. In the strongly ion-solvating environment ofthe DMAC cell, a possible explanation for reaching the end of the dis-charge at only 12 e– out of possible 16 e–/S8 is that as the lower orderpolysulfides are depleted, they become more stabilized in the reducedform and do not readily disproportionate to Li2S and either a higherorder polysulfide or S8 (reactions 5 and 6). Their increased stabilityleads to an increase in the applied overpotential necessary to maintainthe prescribed discharge current and the subsequent end of the dis-charge. In the poorly ion-solvating environment of the DOL-DMEcell, the change in the stability of the lower order polysulfides cannotplay a similar role in reaching the end of the discharge, because almostall of the lower order polysulfides are consumed after a charge transferof only 10 e–/S8. It is possible, however, that the hypothesized forma-tion of S2

2-/Li2S2 intermediate could be contributing to the inability ofthe cell to extract more than 12.6 e–/S8. For example, if the S2

2-/Li2S2

intermediate cannot be further electro-reduced or disproportionated,the conversion of all other species in the cell to S2

2-/Li2S2 and Li2Swill eventually lead to a premature end of the discharge (Figure 4c).

Conclusions

In this work, we introduce a new spectro-electrochemical celldesign with unique characterization features. Typical operando celldesigns offer either a proper electrochemical performance through amodification of a coin, pouch, or Swagelok cell or provide advancedspectroscopic capabilities, such as spectroscopic access to the elec-trodes and the separator for spatially-resolved measurements, but notboth qualities at the same time. In contrast, our cell combines bothexcellent electrochemical performance, as has been demonstrated inthe benchmark tests using a standard Swagelok T-cell, and offers thepossibility to characterize different cell components by pressing theedges of the electrodes and the separator against the X-ray window.Although in this study we use a defocused beam and do not performspatially-resolved measurements, we successfully confirm the spec-troscopic capabilities of the cell by tracking the formation of S3

•-

intermediate and closely matching the previously reported UV-Visspectroscopic results.

After establishing the spectro-electrochemical capabilities of ouroperando cell, we subsequently use it to investigate the influence of thesolvent on the conversion of polysulfides to Li2S by employing cellsbuilt using DMAC and DOL-DME electrolyte solvents. In good agree-ment with previous XRD, electrochemical impedance spectroscopy,and modeling studies, we observe the formation of Li2S after a dis-charge of only 4–5 e–/S8 in both types of cells, which represents theearliest operando XANES detection of Li2S using standard electrolyteand electrode formulations. Furthermore, by offering simultaneousdetection of liquid and solid intermediates, our experiments high-light the advantages of operando XANES over other characterization

Journal of The Electrochemical Society, 162 (7) A1146-A1155 (2015) A1155

approaches, including UV-Vis spectroscopy, HPLC, and XRD, whichcan detect either polysulfides or Li2S but not both species at the sametime.

Additional analysis of the XANES spectra reveals that despite asimilar initial onset of Li2S formation in the two cells, subsequentconversion of polysulfides to Li2S is found to be more rapid in theDOL-DME than in the DMAC cell. Furthermore, at the end of thedischarge, two different compositions are observed in the two cells.In the DMAC cell, we detect the presence of Li2S and a polysulfideintermediate, while in the DOL-DME cell, we detect the presence ofLi2S and what likely is a S2

2-/Li2S2 intermediate. In disagreementwith previous studies, the final spectra do not indicate significantpresence of S8 in either type of the cell. Taken as a whole, the observeddifferences between the operando XANES results obtained in theDMAC and in the DOL-DME cells provide strong evidence that thedisproportionation steps are faster in a poorly rather than a stronglyion-solvating environment, which, as has been recently proposed byLu et al.,14 can explain the higher rate capabilities of Li-S batteriesbased on low dielectric solvents, such as DOL-DME.

Acknowledgments

The described sulfur K-edge XANES measurements were per-formed at the PHOENIX beamline at Swiss Light Source, Paul Scher-rer Institut, Villigen, Switzerland. The authors also thank DiamondLight Source, Didcot, UK for access to beamline I18 (SP 8734) thatcontributed to the results presented here, Konstantin Ignatyev for sup-port during experiments at Diamond Light Source, Anna Eberle andJudith Martin for assessing the discharge behavior of Li-S batteryin electrolytes based on LiClO4, LiTFSI, and other salts, and MarcSchonberger and Erik Faber for technical support in designing andbuilding the operando cell. Portions of the research leading to theseresults have received funding from the European Community’s Sev-enth Framework Programme (FP7/2007–2013) under agreement No.312284 and from the German Ministry of Education and Research(BMBF) under agreement No. 03X4627A. Yelena Gorlin gratefullyacknowledges the Alexander von Humboldt Foundation for her Post-doctoral Fellowship. Authors Yelena Gorlin and Armin Siebel con-tributed equally to this work.

References

1. M. Z. Jacobson, Energ. Environ. Sci., 2, 148 (2009).2. Global EV Outlook: Understanding the Electric Vehicle Landscape in 2020, Interna-

tional Energy Agency (2013).3. EV Everywhere: Grand Challenge Blueprint, US Department of Energy (2013).4. K. G. Gallagher, S. Goebel, T. Greszler, M. Mathias, W. Oelerich, D. Eroglu, and

V. Srinivasan, Energ. Environ. Sci., 7, 1555 (2014).5. B. Scrosati and J. Garche, J. Power Sources, 195, 2419 (2010).6. P. G. Bruce, S. A. Freunberger, L. J. Hardwick, and J.-M. Tarascon, Nat. Mater., 11,

19 (2012).7. J.-M. Tarascon, Phil. Trans. R. Soc. A, 368, 3227 (2010).8. Y. V. Mikhaylik, I. Kovalev, R. Schock, K. Kumaresan, J. Xu, and J. Affinito, ECS

Trans., 25, 23 (2010).9. Fiscal Year 2013 Annual Progress Report for Energy Storage R&D, US Department

of Energy (2014).10. L. F. Nazar, M. Cuisinier, and Q. Pang, MRS Bulletin, 39, 436 (2014).

11. R. D. Rauh, F. S. Shuker, J. M. Marston, and S. B. Brummer, J. Inorg. Nucl. Chem.,39, 1761 (1977).

12. R. D. Rauh, K. M. Abraham, G. F. Pearson, J. K. Surprenant, and S. B. Brummer,J. Electrochem. Soc., 126, 523 (1979).

13. Y. Jung, S. Kim, B.-S. Kim, D.-H. Han, S.-M. Park, and J. Kwak, Int. J. Electrochem.Sci., 3, 566 (2008).

14. Y.-C. Lu, Q. He, and H. A. Gasteiger, J. Phys. Chem. C, 118, 5733 (2014).15. R. Bonnaterre and G. Cauquis, J. Chem. Soc., Chem. Commun., 293 (1972).16. T. Chivers and I. Drummond, Inorg. Chem., 11, 2525 (1972).17. R. P. Martin, W. H. Doub, J. L. Roberts, and D. T. Sawyer, Inorg. Chem., 12, 1921

(1973).18. X. Ji and L. F. Nazar, J. Mater. Chem., 20, 9821 (2010).19. N. A. Canas, S. Wolf, N. Wagner, and K. A. Friedrich, J. Power Sources, 226, 313

(2013).20. S. Walus, C. Barchasz, J.-F. Colin, J.-F. Martin, E. Elkaim, J.-C. Lepretre, and

F. Alloin, Chem. Commun. (Cambridge, U. K.), 49, 7899 (2013).21. L. Yuan, X. Qiu, L. Chen, and W. Zhu, J. Power Sources, 189, 127 (2009).22. K. Kumaresan, Y. Mikhaylik, and R. E. White, J. Electrochem. Soc., 155, A576

(2008).23. L. A. Huff, J. L. Rapp, J. A. Baughman, P. L. Rinaldi, and A. A. Gewirth, Surf. Sci.,

631, 295 (2015).24. N. A. Canas, D. N. Fronczek, N. Wagner, A. Latz, and K. A. Friedrich, J. Phys.

Chem. C, 118, 12106 (2014).25. C. Barchasz, F. Molton, C. Duboc, J.-C. Lepretre, S. Patoux, and F. Alloin, Anal.

Chem., 84, 3973 (2012).26. A. Kawase, S. Shirai, Y. Yamoto, R. Arakawa, and T. Takata, Phys. Chem. Chem.

Phys., 16, 9344 (2014).27. M. Cuisinier, P.-E. Cabelguen, S. Evers, G. He, M. Kolbeck, A. Garsuch, T. Bolin,

M. Balasubramanian, and L. F. Nazar, J. Phys. Chem. Lett., 4, 3227 (2013).28. M. U. M. Patel, I. Arcon, G. Aquilanti, L. Stievano, G. Mali, and R. Dominko,

ChemPhysChem, 15, 894 (2014).29. T. A. Pascal, K. H. Wujcik, J. Velasco-Velez, C. Wu, A. A. Teran, M. Kapilashrami,

J. Cabana, J. Guo, M. Salmeron, N. Balsara, and D. Prendergast, J. Phys. Chem. Lett.,5, 1547 (2014).

30. M. E. Fleet and X. Liu, Spectrochim. Acta Part B, 65, 75 (2010).31. M. Cuisinier, C. Hart, M. Balasubramanian, A. Garsuch, and L. F. Nazar, Adv. Energ.

Mat., 1401801 (2015).32. M. Cuisinier, P. E. Cabelguen, B. D. Adams, A. Garsuch, M. Balasubramanian, and

L. F. Nazar, Energ. Environ. Sci., 7, 2697 (2014).33. J. Gao, M. A. Lowe, Y. Kiya, and H. D. Abruna, J. Phys. Chem. C, 115, 25132

(2011).34. Q. Pang, D. Kundu, M. Cuisinier, and L. F. Nazar, Nat. Commun., 5, 4759

(2014).35. M. A. Lowe, J. Gao, and H. D. Abruna, RSC Advances, 4, 18347 (2014).36. H. Jha and H. A. Gasteiger, DE Patent No. 102013005082 A1. Munich, Germany:

Deutsches Patent- und Markenamt (2014).37. A. M. Flank, G. Cauchon, P. Lagarde, S. Bac, M. Janousch, R. Wetter,

J. M. Dubuisson, M. Idir, F. Langlois, T. Moreno, and D. Vantelon, Nucl. Instrum.Methods Phys. Res., Sect. B, 246, 269 (2006).

38. B. Ravel and M. Newville, J. Synchrotron Rad., 12, 537 (2005).39. G. Brauer, Handbuch Der Praparativen Anorganischen Chemie, Enke Verlag,

Stuttgart (1975).40. P. T. Cunningham, S. A. Johnson, and E. J. Cairns, J. Electrochem. Soc., 119, 1448

(1972).41. G. N. George, M. Gnida, D. A. Bazylinski, R. C. Prince, and I. J. Pickering, J. Bac-

teriol., 190, 6376 (2008).42. I. J. Pickering, R. C. Prince, T. Divers, and G. N. George, FEBS Lett., 441, 11 (1998).43. G. Almkvist, K. Boye, and I. Persson, J. Synchrotron Rad., 17, 683 (2010).44. H. Schneider, C. Gollub, T. Weiß, J. Kulisch, K. Leitner, R. Schmidt,

M. M. Safont-Sempere, Y. Mikhaylik, T. Kelley, C. Scordilis-Kelley, M. Laramie,and H. Du, J. Electrochem. Soc., 161, A1399 (2014).

45. Y.-C. Lu, E. J. Crumlin, T. J. Carney, L. Baggetto, G. M. Veith, N. J. Dudney, Z. Liu,and Y. Shao-Horn, J. Phys. Chem. C, 117, 25948 (2013).

46. T. Gross, L. Giebeler, and C. Hess, Rev. Sci. Instrum., 84, 073109 (2013).47. S. J. Harris, A. Timmons, D. R. Baker, and C. Monroe, Chem. Phys. Lett., 485, 265

(2010).48. Y.-L. Ding, B. M. Goh, H. Zhang, K. P. Loh, and L. Lu, J. Power Sources, 236, 1

(2013).49. M. E. Fleet, Can. Mineral., 43, 1811 (2005).